[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
RE: [cdt-dev] New project model

Hi Dave,

There are two aspects of migration/backward compatibility support we are
considering.

1. Existing projects support
2. Existing integration support.

1. Existing project support: the existing projects must be supported.
When the project is loaded the user is asked whether the project should
be converted. If the answer is yes, the project is converted and stored
in the new format, otherwise the project information is converted, but
not stored and will be available in the read-only mode.
We have introduced a new extension that is intended to perform a
conversion.
We already have the old Managed make projects successfully converted to
the new format and are planning to implement the converter for the
standard make projects soon as well.
2. Existing integration support. It is planned to support existing
integration as much as possible, although there might be some problems
with providing the 100% backward compatibility. I was mentioning those
problems in the New Project Model Core design (see the "Backward
Compatibility" section at the end of the document), but unfortunately
received no feedback on that.

>From the Path Entries point of view the question is does anyone needs
this functionality to be remained in the CDT. If so it can be supported
as a wrapper around the new functionality. Note that the path entries
API will work only for one configuration (used by the CModel and
indexer) and will not provide the per-language settings granularity i.e.
will collect/store the settings from/to all language settings defined
for the configuration.

>From the Std Make Scanner discovery profile mechanism, it will remain
unchanged and will be utilized by the CDT Build System for discovering
includes/symbols settings as it is done now.

Mikhail

 
-----Original Message-----
From: cdt-dev-bounces@xxxxxxxxxxx [mailto:cdt-dev-bounces@xxxxxxxxxxx]
On Behalf Of David Inglis
Sent: Friday, December 01, 2006 10:18 PM
To: CDT General developers list.
Subject: RE: [cdt-dev] New project model

Question. Is a migration strategy included in this work for converting
old
project to use the new model?  An example is we have implemented
extensions
to the path entries framework where we dynamically contribute path
information to project (std. make does is as well with its discovery
mechanism).

> -----Original Message-----
> From: cdt-dev-bounces@xxxxxxxxxxx [mailto:cdt-dev-bounces@xxxxxxxxxxx]
On
> Behalf Of Sennikovsky, Mikhail
> Sent: Friday, December 01, 2006 6:47 AM
> To: CDT General developers list.
> Cc: Brunauer, Walter
> Subject: RE: [cdt-dev] New project model
> 
> Hi Markus,
> 
> The New Project Model work is in progress now. Most of the new
> functionality is implemented. Since most of the new Project Model
> features depend on each other and affect different areas of
> functionality we can not commit them in peaces to ensure CDT integrity
> and stability.
> 
> We are currently planning to perform the major commit containing most
of
> the primary functionality in the end of January 2007.
> Since the commit is going to be quite big, we are definitely going to
> post patches for review/testing to the New Project Model bugzilla
before
> committing them.
> 
> To bring some clarity to the integration process of the Core New
Project
> Model with other core functionality here are some of my thoughts
> regarding this:
> 
> My plan is to make the integration of the New Project Model with other
> core functionality as painless as possible to avoid merge conflicts
and
> reduce amount of work necessary for integration. From the indexer
point
> of view all "information provider API" used by the indexer currently
> (LanguageManager, ScannerInfoProvider, etc.) will be kept and will be
> used in the normal way:
> - Since the new project model will contain the new per-language
> Includes/Symbols settings functionality that will replace the current
> path entries framework, the new ScannerInfoProvider is implemented
that
> provides the scanner info based upon the New Project Model settings.
> - since the New Project Model will allow language settings
customization
> (i.e. in case more than one languages are associated with one content
> type, the New Project Model will allow specifying the language to be
> used).
> I have added few new methods to Language manager for getting the
> language e.g. given an IFile, etc. to allow obtaining the necessary
> language for the file. I have also modified the LanguageManager
clients
> to use these new methods.
> 
> Please let me know if you have any questions or concerns about the
dates
> or integration details. We could discuss them by phone if necessary
> (e.g. on the next Indexer call)
> 
> Regards,
> Mikhail
> 
> 
> -----Original Message-----
> From: cdt-dev-bounces@xxxxxxxxxxx [mailto:cdt-dev-bounces@xxxxxxxxxxx]
> On Behalf Of Schorn, Markus
> Sent: Friday, December 01, 2006 12:16 PM
> To: CDT General developers list.
> Cc: Brunauer, Walter
> Subject: [cdt-dev] New project model
> 
> Hi Mikhail!
> I am wondering what the current progress on the new project model is.
> I'd like to understand when you are planning to commit the changes to
> CVS.
> Is it going to be a all-in-one commit (which milestone?), or are you
> planning a couple of independent steps? Can we get early access to
> patches, so that we can try out things up front?
> 
> Markus.
> _______________________________________________
> cdt-dev mailing list
> cdt-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/cdt-dev
> _______________________________________________
> cdt-dev mailing list
> cdt-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/cdt-dev
_______________________________________________
cdt-dev mailing list
cdt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/cdt-dev